ChatGPT Shopping Gets Visual Browsing and Product Comparisons

OpenAI has expanded ChatGPT’s shopping experience with richer product discovery, including visual browsing, side by side comparison, image-led shopping, and broader merchant data coverage. The March 24 rollout, detailed in Powering product discovery in ChatGPT, reaches ChatGPT Free, Go, Plus, and Pro users this week. For developers building commerce agents, shopping assistants, or merchant integrations, the important change is architectural: OpenAI is turning the Agentic Commerce Protocol into a discovery layer, not just a checkout handoff.

Product Experience

The user-facing update is straightforward. ChatGPT can now present products visually, compare options across fields like price, reviews, and features, and respond better to image-based shopping requests where a user uploads a photo and asks for similar items.

OpenAI also says results are faster, more relevant, and more current. Those improvements matter because shopping UX fails quickly when catalog data is stale or product normalization is weak. If your application depends on AI-assisted product search, the retrieval problem is no longer just “find relevant text.” It is “resolve the right SKU, attributes, seller, and current offer,” which is closer to structured retrieval than classic RAG. This is the same shift behind recent work on context engineering and production RAG systems.

Strategy Shift

The bigger change is OpenAI’s move away from Instant Checkout as the center of the commerce story.

In September 2025, OpenAI positioned “buy it in ChatGPT” around native checkout, starting with U.S. Etsy sellers and promising more than 1 million Shopify merchants to follow through the same general path. On March 24, the emphasis changed to discovery inside ChatGPT and transaction completion on merchant-owned websites, apps, or merchant-specific ChatGPT apps.

OpenAI’s merchant flow now centers on discovery, structured catalog sharing, and merchant-controlled conversion. Purchases are completed on merchant properties rather than through a standalone universal checkout layer inside ChatGPT.

ACP as Merchant Data Infrastructure

This release extends ACP into the connective layer between merchants and users throughout the discovery flow. Merchants can provide product feeds and promotions, and OpenAI supports delivery paths through providers including Salesforce and Stripe.

That matters because ACP is now functioning more like an interoperability protocol for AI-native commerce. If you build agents, this resembles the broader trend toward standardized tool and data interfaces, similar in spirit to how developers think about MCP or structured function calling. The difference is domain specificity. ACP is aimed at catalogs, merchant metadata, and downstream purchase flows.

OpenAI says ACP will later support additional features including personalization, local availability, and ETAs. Those are all high-value retrieval and ranking features, because they directly affect conversion quality.

Merchant Paths

OpenAI currently exposes three practical ways for merchants to show up in ChatGPT shopping flows.

Path	What it enables	Current detail
Search crawling	Baseline discovery	Merchants can appear if OAI-SearchBot is allowed
Direct feeds and partner catalog integrations	Better freshness and accuracy	Includes Shopify Catalog and direct feed applications
ChatGPT apps	Branded, deeper shopping experiences	Highlighted for larger merchants such as Walmart

This is a more realistic platform design than a one-size-fits-all in-chat checkout. Large retailers want account linking, loyalty systems, merchant payments, and branded environments. Smaller merchants want broad discovery with minimal integration work.

Retailer Coverage

OpenAI names seven retailers already integrated into ACP for discovery: Target, Sephora, Nordstrom, Lowe’s, Best Buy, The Home Depot, and Wayfair.

For Shopify merchants, the baseline path is especially notable. OpenAI says Shopify Catalog is already integrated into ChatGPT, and individual merchants do not need additional work for basic representation. Shopify has separately described that catalog infrastructure as operating across billions of products, which helps explain how OpenAI can expand product coverage quickly without bespoke feeds for every merchant.

Walmart as the Model for Deeper Integration

The clearest same-week implementation is Walmart’s in-ChatGPT app experience. OpenAI says it supports account linking, loyalty, and Walmart payments, and is available now on the web, with iOS and Android support coming shortly.

This is the practical template for enterprise merchants. Discovery happens in the shared ChatGPT surface. High-intent shopping then moves into a merchant-controlled environment with its own identity, payment, and retention mechanics. If you work on AI agents vs chatbots, this is a useful line: the assistant handles discovery and orchestration, while the merchant app handles the transactional system of record.

Ranking and Selection

OpenAI says product results are selected independently, are not ads, and are not influenced by partnerships. Selection uses the user’s query and context, structured metadata from first-party and third-party providers, model reasoning, and product and safety policies.

Merchant ranking on product detail pages can consider availability, price, quality, whether a merchant is the maker or primary seller, and whether Instant Checkout is enabled. OpenAI also notes that titles and descriptions may be simplified from merchant data, and that price or shipping information can lag recent merchant-side changes.

If you build merchant tooling for this ecosystem, structured catalog quality now matters as much as prompt quality. Normalize attributes, keep inventory and price feeds fresh, and decide whether you need baseline discovery, direct feeds, or a full ChatGPT app.